A determination of the low work function planes of LaB6

Surface Science 146 (19R4) W-599 North-Holland, Amsterdam

A DETERMINATION

583

OF THE LOW WORK FUNCTION

PLANES

OF

La% M. GESLEY Oregon

Received

and L.W. SWANSON

Graduate Center, 19600 N. W. Walker Roud, Beauerlon, Oregon 97006, USA

10 April 1984; accepted

for publication

2 July 1984

A study of the temperature and orientation dependence of the field ion and electron emission and the average work function for a La& emitter have been carried out between 77 and 1800 K. These results indicate that the (310) surface is thermally stable and has the lowest work function of any single crystal plane of LaE&. This result is confirmed by field emission retarding potential (FERP) work function (+) measurements at 300 K which gives +(310) = 2.50 eV. The results show that the relative orientation dependence of the room temperature work function is +(310) < +(210) c +(lOO) i +(llO) < $(lll) -C cp(211). This ordering is in accordance with the relative dipole contribution to the work function when surface reconstruction is included. The orientation dependence of the field electron emission is sensitive to field evaporation, thermal equilibration, and carbon contamination.

1. Introduction In the development of a high brightness electron source considerable attention has been focused on the crystallographic dependence of the work function of La$. For the low index planes the orientation dependence of the work function has been determined by thermionic emission (TE) [I,Z], field emission retarding potential (FERP) [l], and photoelectric (31 methods to be +(lOO) -C(~(110) c cp(111). Many studies have focused on the low index planes due to their thermal stability and ease of fabrication. Field-ion microscopy (FIM) has presented direct visual evidence of this stability on an atomic scale [1,4]. However severa studies have indicated that certain high index planes may exhibit even lower work functions and thermal stability (i.e., stability with respect to macroscopic surface reconst~ction). A TE study measured a 2.41 eV work function for the (346) plane [l]. An ultraviolet photoelectron study (UPS) measured a 2.2 eV work function for the (210) plane [S]. Another TE study found +(210)< (p(IO0) [2]. In this study field emission microscopy (FEM) and FIM studies of La$ have been carried out as a function of temperature of investigate the possible existence of high index, low work function planes which are stable at high 0039-6028/84/$03.~ 0 Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)

temperatures. The combined use of FIM and FEM allows one to qualitatively discern the crystallographic dependence of work function from local electric field variations. Auger electron spectroscopy (AES), FERP. and TE measurements on macroscopic single crystals have also been performed in order to verify that the relative ordering of work function is: +(310) i +(210) < +(lOO) < (p(110) < +(111) < #(211). Measurements of the work function and FEM patterns of an originally field evaporated emitter as a function of annealing temperature, and studies on macroscopic samples using FERP and AES. indicate that low work function surfaces generally correspond to a higher surface concentration of lanthanum atoms. In a similar study concerning the temperature dependence of the FIM and FEM patterns carried out by Futamoto et al. [4] symmetric field emission patterns were only obtained in the range of 1700 -Z T < 1950 K. To obtain reproducible and coherent FEM patterns over the entire range of temperatures requires a combination of high vacuum ( ( lop9 Torr) and beginning with a low temperature field evaporated surface. Our results indicate that the FEM pattern is sensitive to certain ambient gases and that raising the temperature of the crystal to 1900 K is an insufficient method for removing surface carbides.

2. Experimental The macroscopic sampies of La$ were obtained from a single crystal rod (1 .O mm diameter) prepared by a float zone melting technique described by Gibson and Verhoeven [6]. Single crystal stoichiometry was determined by chemical analysis. Crystal mounting procedure and heating configuration have been explained previously [l]. Measurements made on macroscopic single crystal surfaces were carried out in an IJLTEK TNB-X 250 l/s ultra-high vacuum system containing a Physical Electronics CMA Model lo-155 Auger Electron Spectrometer, FERP gun for room temperature work function measurements, and an outgassable guard ring anode for obtaining thermionic work functions. The AES measurements were performed with a 10 PA primary beam current with 5 kV beam energy. Quantitative analysis of the AES data used the peak-to-peak heights of the derivative of the N(E) curve with one volt modulation. The thermionic (&) and FERP (rpr) work function measurements have been described in detail elsewhere [1,7]. Both AES and work function measurements were carried out after heating the crystal in excess of 120 h at 1800 K. The FIM and FEM studies were performed in a standard, low temperature (77 K) microscope using a microchannel plate to intensify the hydrogen or helium ion image and the field electron image. The bakeable, ultra-high vacuum system had a base pressure of 5 X lo-“’ Torr. Image gas pressures for FIM studies were typically 1 x 10e4 Torr and all FEM work was performed

M. Gesley, L. W. Swanson /

Low work function

planesof LoBe

585

below 1 x 10m9 Torr. Field evaporation of La$ in the presence of H, occurs slowly at 220 MV/cm which is near the best image field (BIF) for H, image gas. In contrast,

the field evaporation

of LaB, in the presence

of a noble gas or

in vacuum occurs at 420 MV/cm. The microscopy studies employed needle-like La$ the molten Al solvent method described by Futamoto

crystals obtained from et al. [8]. The crystals

were mounted

with TaC

between

acted as an adhesive. rods. Temperature [l]. During literature

measurements

ribbons

apparent

nants could drastically possibility the emitter temperature

has

powder

were made pyrometrically

as described

study and subsequent

that small concentrations

analysis

of surface

that

support earlier of the

contami-

alter the emission distribution. In order to avoid this was first heated to 1650 K (P < 1 X lo-’ Torr). This

been

CO, H,O,

coated

were then spot welded to tungsten

the course of the microscopy it became

adsorbed

two rhenium

The ribbons

determined

and 0,

by AES

from the emitter

to be high

enough

shank and support

to remove assembly

[l].

It was then field evaporated to remove any compounds remaining on the surface of the emitting area. The starting point for all of the FIM and FEM experiments was thus an emitter with a field evaporated surface free of any shank

contaminants

that could

migrate

to the emitting

area during

higher

temperature annealing. Adsorption of hydrogen on a He-field evaporated emitter has been found to have negligible effect on the electron emission properties [l]. Two additional precautions were taken to avoid any possible effect hydrogen might have. First, whenever consecutive FIM, FEM images were recorded the field was never removed

until the pressure had been reduced below 10m9 Torr, after which the

field electron emission measurement was then promptly electron emission properties as a function of temperature obtained concluded

when

no field

that hydrogen

ion

microscopy

adsorption

recorded. coincided

was performed.

Second, the with results

Therefore

had no effect on the experimental

it was results.

3. Results 3. I. FEM/

FIM results

Beginning with a clean, field evaporated LaB, emitter FIM and FEM images were obtained at 77 K along with a measurement of the total field emission

current

sample for two Fowler-Nordheim

(I)

as a function

of applied

voltage

(V)

after

heating

minutes in 100 K increments. The slope (m) (FN) plot of the I(V) characteristic is given by

m = d[ln(1/V2)]/d(l/V)

= bG3’*/jl,

where + is the average emitter work function,

of

the the

(1) p = F/V,

F is the electric field at

the emitter surface and h is a constant. The variation in FN slope with annealing temperature is shown in fig. 1. The error bars correspond to the variation between two experimental runs. The best image voltage (BIV) for the H, FIM image at 77 K, was constant over the range of emitter annealing temperatures implying that j3 in eq. (1) remained unchanged during the course of the temperature processing. By expressing the electric field in the approximate form F=

V/kR,

0.)

where R is the emitter radius and k is a geometrical factor with values between 4 and 8 depending on the emitter shape 191, then from eqs. (1) and (2), /3 = ( kR )- i. The relationship I’,,,, a R for a constant F is verified in fig. 2 where the emitter radius was calculated by the ring counting method [9] between the (110) and (210) planes for several different emitter radii. Given that p and hence R remained constant during the course of the temperature processing, the changes in the FN slope must be interpreted as variations in the average emitter work function. For each point on the graph in fig. 1 corresponding FIM and FEM images

1 -60

0

0

-

-4o-

W -30

-’

: cn z ’ CL

-2o-

-I 0 -

ool

0

”

” 200

” 400

” 600

800

”

TEMPERATURE

IO00

”

1200

”

1400

i

”

1600

1800

(K)

Fig. 1. Plot shows the variation of the FN slope of a La& emitter (after initial field evaporation H, at 77 K) with 2 min heating intervals at the indicated temperatures.

in

M. Gesley, L. W. Swanson / Low work function planes of LaB,

587

were obtained. Representative micrographs, which were reproducible within a range of 100 K, are shown in fig. 3. Beginning with a Hz-field evaporated surface shown in fig. 3a (3a’) no significant change was found in the FEM image until approximately 725 K, and not in the FIM pattern until 1000 K. Fig. 1 indicates that for T < 800 K the average work function increases with temperature. The fact that the FIM image remains unchanged with respect to the 77 K field evaporated surface indicates that La and B diffusion are inconsequential for T-c 1000 K and that the initial increase in the average work function with temperature must be attributed to local surface atom rearrangement. Fig. 3b’ shows that heating for 2 min at 910 K results in enhanced electron emission from the { 110) regions. The average work function begins decreasing at this point although it still exceeds the value for the original evaporated surface at 77 K. At 1295 K the (110) planes are no longer resolved as shown in fig. 3c. The electron emission becomes more uniform (see fig. 3~‘) and the average work function undergoes a rapid decrease. Heating the sample in the range 1355 < T < 1770 K results in a dramatic change in the FEM image (see figs. 3d to 3f) and a sharp reduction in the average work function. In addition, the FIM patterns show a growth of the {ill}, {llO}, (112) and to some extent the (100) planes. As the temperature increases towards the upper part of the stated temperature interval bright emission, initially observed from the (210) planes, proceeds to the (310) planes. Evidence indicating that ~(310) < ~(210) can be obtained from fig. 3f by carefully noting the positions of the planes from the FIM patterns. Field enhancement is centered on the (210) planes as shown by fig. 3f whereas the enhanced electron emission is centered more on the { 310) planes. Although the II3

I

,

I

I

I

I

I

I

I 700

I 600

16 14 3

IZIO -

> m

6-

)

6420 0

I 100

I 200

I 300

I 400

I 500

RADIUS

Fig. 2. Graph of H, BIV (kV) versus La& emitter counting between (100) and (210) planes.

I 600

900

(8,

radii.

Emitter

radii calculated

by FIM ring

588

M. Gesfey, L. W. Swanson / Low work function planes of LaB,

(a)

W

T=77K

T= 910K

(b’)

T= l295K

(d)

T=l355K

kV for the hydrogen Fig. 3. FIM funprimed) and FEM (primed) Ftatterns. The BIV=13 to a point on fig. 1. images. The patterns at the indicated tempera1 tores correspond

ic

M. Gesley, L. W. Swanson

/ Low work funcrion

planes of LB6

591

FIM image fig. 3f shows some field enhancement in the {210}-(310) regions, the BIV was unchanged indicating the average value of emitter radius remained constant throughout the annealing process. Thus the large decrease in the FN slope at T > 1355 K must be attributed to a reduction in @Iwith the (310) planes exhibiting the lowest work function for the thermally annealed end form. Using a hemispherical thermionic emission microscope, Shimizu et al. have observed bright emission from the (210) planes at - 1350 K [lo]. This is consistent with the present field emission results which indicate that the transition in the low work function planes, i.e. from (210) to {310}, occurs between 1355 and 1410 K (see figs. 3d’ and 3e’). Several previous studies have reported that enhanced field emission occurs in the (211) region (fig. 6d [l], fig. 2a [4] and fig. 4 [ll]) for an emitter heated at T - 1800 K. Our studies show that this emission distribution is obtained by heating in the presence of hydrocarbons. Oxygen was ruled out as a contaminant as its compounds would be removed by heating at this temperature, carbon however is not [l]. Only by field evaporation of the emitter followed by heating in high vacuum (< lop9 Torr) could we reproducibly obtain the fig. 3f’ pattern. Comparing FN slopes from the carbon contaminated emitter, which resulted in bright emission from the (211) planes, with the FN slope corresponding to the (310) emission shows: m(211)/m(310) = 1.3. This is consistent with FERP and AES measurements which indicate that a small amount of surface carbon significantly increases the work function [I]. 3.2. Macroscopic

single crystal

results

Room temperature FERP work function measurements on macroscopic single crystals, summarized in table 1, are consistent with the present field emission results which indicate that +(310) < +(210) += (~(211). In fact the Table 1 Work function and AES summary for the indicated crystal primary beam energy 5 kV; relative dipole contributions crystal faces are indicated

faces and stoichiometries of LaB,; AES A+o to the work function of various

Plane

AGo (n =I)

B(179)/La(625) (T=3OOK)

+,(300 K) (eV)fO.OS

&(1600 K) (eV)*0.04

LaBe.cs(310) LaB,.e,(210) La$.es(LOO) LaB5.s6(100) LaBs.s,(I 10) LaBs.s,(LlL) LaB,.,(211)

_ - 11.38 ‘) -6.48 - 10.94 - 10.94 - 6.05 a) 0.00 0.00 a) + 7.34 _

4.8 4.1 4.4 4.6 5.8 6.1 5.1

2.50 2.55 2.69 2.60 2.65 2.90 3.05

2.42 2.46 2.57 2.52 2.64 2.90 2.92

‘) Based on reconstructed

surfaces

M. Gesley, L. W. Swanson / Low work function planes of LaB,

592

(211) plane appears to have one of the highest work functions of all measured planes. The TE work function measurements are generally 0.1 eV less than the FERP values for a given plane. Fig. 4 shows the temperature variation of the effective TE work functions for the (loo), (210) and (310) planes. The (p, values for the (310) plane, although taken over a limited temperature range, are slightly lower than the c+~values for the (210) plane. For both the (310) and (210) planes no change in work function was observed upon heating at 1800 K for 120 h. In another experiment a La$ crystal was cleaved along a (100) plane in ultra-high vacuum and examined by AES. Table 2 summarizes the AES results for the cleaved and annealed La$(lOO) surfaces.

2.4

2.2 1;

I

0

1300

I

I

1400

I500

I

1600 T

Fig. 4. Variation of the thermionic (210) crystal faces.

I

I700

I

1800

I

10

(K)

work functions

& with

temperature for the (100). (310) and

M. Gesley,L. W. Swanson / Low work

functm planes of LAB,

593

4. Discussion An

indication

obtained

of the nature

by considering

of the Hz-field

evaporated

surface

can

be

the slope of the graph in fig. 2. The value of the slope

(equal to kF) coupled with an estimate of the hydrogen best image field (BIF) of 220 MV/cm [9] results in a geometrical factor k = 6. This result is obtained by assuming

the observed step height employed in the ring counting method is equal to the lattice constant a, = 4.14 A of LaB,. Using a step height value of a,/2 implies k = 12 whereas a value of 2a, results in k = 3, both of which are outside the typical range of values for the geometrical factor. Since the net plane rings observed in the FIM correspond to ledge atoms, a step height equal to the lattice constant implies that only one atomic species exists at the ledge sites (except for a nonpolar plane such as an unreconstructed (110)). This conclusion was also arrived at in an earlier FIM study using noble gases to image rare-earth

hexaborides

[12].

A question of interest is, which element is being imaged by FIM? Futamoto and Kawabe [12] argue that the FIM image of a field evaporated surface is a result

of La

ionization

atoms

probability

only.

Their

is greater

argument

is based

on

the idea

that

the

above the La atom due to the charge transfer

from the La atom to the $ octahedra.

But this depends on the assumption

that

either the $ octahedra and the La atom exist in equivalent kink or ledge sites or that lanthanum is outermost on the low index planes of the field evaporated surface. The first possibility cannot hold geometrically for the (100) plane, one which is visible in the FIM for both hydrogen and helium field evaporated endforms [l]. The second possibility contradicts both the results of atom probe studies on surfaces field evaporated with noble gases [13] and the results presented here which conclude that a field evaporated La$(lOO) surface has boron atoms outermost. Although Futamoto and Kawabe 1121 recognized this possibility,their conclusion requires the field interacting with a La atom 2.1 A below the surface to be comparable to the surface field interacting with the boron octahedra. However, the penetration depth of the electric field for a free electron model using the Thomas-Fermi approximation [9] is only 0.7 A when applied to La$. This significantly reduces the ionization probability above the La atoms and hence they must have a negligible effect on the FIM image of the (100) plane. In a detailed study of the field evaporation of the La$(lOO) plane in the presence of noble gases carried out by Murakami et al. [13] only boron was observed

to field evaporate

from the central

region of the net plane,

whereas

both La and B field evaporated at the net plane edge. The authors concluded that field evaporation in noble gases or vacuum proceeds by decomposition of the boron octahedra with partial removal of boron atoms followed by simultaneous evaporation of the lanthanum atom and the remaining boron atoms from the decomposed octahedra. Thus, upon field evaporation the polar (100)

M. Gesley,

594

L. W. Swanson

/ Low work function

planes

of

LOB,

face always results in a configuration with boron outermost, The alternative possibility of having lanthanum outermost, caused by initial evaporation of lanthanum ions with simultaneous decomposition and partial removal of boron atoms from the $ octahedra, then followed by evaporation of the remaining boron atoms, is unlikely due to the difference in the boron and lanthanum field evaporation rates. It has been reported that at the stoichiometric composition the sublimation energies of B and La are equal [14]. Thus, one can show from the image hump field evaporation model [9] that under conditions of uniform work function and field strength that the difference in activation energies Q for field evaporation of La and B is Q, - QL, = I,+ - I,,,+ = 2.68 eV,

(2)

where I,+ and I,,+ are the respective first ionization potentials. This implies that in vacuum or in the presence of noble gases, La will be field evaporated at a much higher rate than B from an initially thermally equilibrated surface. The evaporation field is significantly reduced by H,, as discussed previously. presumably through boron hydride formation [15]. Nevertheless, in view of the near equality of the BIV, FN slopes and hence, work function [l] for the emitter field evaporated in either H, or He image gas environments, we conclude that both surfaces are similarly boron rich and that the observed step heights used in fig. 2 are due to the presence of boron in the ledge sites with the possible exception of the nonpolar (110) planes where both lanthanum and boron may exist. The present experiments indicate that field evaporation generally leaves a boron rich surface and that subsequent heating at T - 1800 K results in a low work function (310) surface caused by diffusion of La atoms to the surface either from the emitter shank or from the underlying bulk without a large by three change in the average emitter radius. This model is supported independent experiments. First, a comparison of the AES result of the fractured and annealed (100) surfaces summarized in table 2 shows that the La(78 eV)/La(626 eV) ratio increases upon heating at 1700 K. Because the 78 eV Auger electron has a shorter escape depth than the 625 eV electron the increase in the

Table

2

Auger

peak

energy

5.0 kV and T=

height

Fractured Annealed

(1700

K)

ratios

for in situ fractured

and

annealed

La$(lOO)

crystals;

primary

300 K B(179)/La(78)

B(179)/La(625)

La(78)/La(625)

3.7

5.2

1.4

1.0

4.6

4.6

beam

M. Gesley, L. W. Swanson / Low work function

planesof LaB,

595

La(78 eV)/La(625 eV) ratio implies an increase in the surface La concentration upon annealing. If we assume the surface structures of the field evaporated and fractured surfaces are similar and represent a nonequilibrium surface, i.e. no surface reconstruction, then it can be concluded that the large reduction in m (or +) observed in fig. 1 at - 1400 K is due to an increase in surface lanthanum concentration. Secondly, in a time of flight atom probe study a marked increase in the La/B ratio occurred when heating an originally field evaporated emitter at or above 1400 K for 5 min in vacuum [16]. As shown in fig. 1, this also is the temperature where the average emitter work function undergoes a large decrease. Finally, assuming the field evaporated surface has a work function equal to elemental boron +u = 4.6 eV [17], one may estimate the work function of the (310) surface from the slope of the FN graph for the field evaporated and annealed (1800 K) surfaces using $(310)

= +a [m(310)/m(evap)12”.

This results in ~(310) - 2.1 eV. Although the TE and FERP measurements of +(310) are - 0.4 eV higher than this estimated value, the calculation does show that the work function change between the field evaporated and annealed surface is large and on the order of 2.3 + 0.2 eV and supports the idea that the field evaporated surface is boron rich. Binary systems with cubic lattice structure, such as the rare-earth hexaborides, possess certain crystal faces, e.g. (210) (111) and (100) that are polar in nature and others, e.g. (llO), which are non-polar provided surface reconstruction does not occur. Angle-resolved photo-electron spectroscopy (ARXPS) has also shown that for annealed LaB, surfaces the La atom is outermost on the (100) face, and that the (111) and non-polar (110) faces are reconstructed by normal displacements of the surface lanthanum by 1.2 and 1.78 A respectively [3]. The unreconstructed and reconstructed surfaces are depicted in fig. 5 for the low index planes of LaB,. According to the latter picture the La atom is outermost on the (100) and (110) planes and parallel with the Bb octahedra on the (111) face. Thus, the separation between the outermost lanthanum layer and the center of the $ plane is 2.1, 1.6, and 0.0 A for the (100) (110) and (111) planes respectively. From table 1, which summarizes the work function and AES results for various crystal faces, one generally finds that the relationship between work function values and B/La ratios (based on the B(KLL)/La(MNN) AES ratio) is such that (excepting the (211) and (310) planes) the work function decreases as the surface lanthanum concentration increases. Conclusions concerning the relation between the true surface concentration of lanthanum and boron and the measured AES ratios must be accepted with some caution as it is known that the latter are dependent on the Auger

M.

596

Gesley,

L. W. Swanson / Low work function planes of LOB,

electron’s exit angle [18], which may explain why the (211) and (310) planes exceptions to the above-stated relation. Nevertheless the relative ordering work function for the low index planes can be rationalized by considering variation in the surface dipole contribution to the work function due to position of the electropositive La atoms relative to their nearest neighbor octahedra as seen in fig. 5. This question can be addressed more quantitatively by assuming that dipole moment due to the La-B chemical bond dominates the variation in outer part of the work function A+, with crystal face and is given by

where uLa is the surface atom density moment k per surface La atom is

of the outermost

are of the the $ the the

La layer and the dipole

J p = 7

c

d, cos 8,.

l-1

The dipole moment normal to the surface is obtained from eq. (5) by considering the j nearest neighbor Bb octahedra as a point negative charge of value ne where d, is the La-$ distance and 8, the angle of the La-$ bond

[II01 t

[OIOI t

Ei J2j

l.2a Reloxatlon

t

1200a Relaxotlon

0 .,D

1 Fig.

5. Sideview

proposed

of (100).

(IlO),

vertical displacement

(111).

II

La IONS B

IONS

B-B BONDS] and (210)

faces of La&.

Dashed

circles

of La atoms due to surface layer reconstruction.

represent

the

M. Gesley,L. W. Swanson / Low work

funciron planesof LOB,

591

with the surface normal. Thus, if one assumes that the relative displacements of the La atoms given by ref. [3] are correct, it can be see from table 1 that ~(100) < +(llO) -C +(lll) is the order predicted by eqs. (4) and (5). If we momentarily assume there is no reconstruction of the (210) plane and compare AC+, for the (210) with the corresponding values for the (loo), (llO), and (111) planes (reconstruction included), it is found from eq. (4) that the orientation dependence of the dipole contribution is A+,(lOO) < A+o(210) < A+,(llO) < A+,(lll), assuming unit charge transfer per dipole, i.e., n = 1 in eq. (5). Reconstruction of at least some of the lanthanum atoms on the (210) plane is suggested by ARXPS results that indicate La atoms are outermost on the (210) and that they float slightly out of their positions [5]. In addition, LEED results show a (1 x 1) pattern for the (210) indicating the surface atoms are in registry with the bulk [S]. A proposed reconstruction consistent with the above data can be obtained by moving the La atom nearest the $ ledge along the [120] direction to the plane intersecting the $ octahedra on the (210) plane (see fig. 5). This results in an increased dipole moment per unit area. The orientation dependence of the surface dipole contribution as given in table 1 is then A+,(210) < A+,(lOO) -CA+,(llO) < A+,(lll) which is consistent with the orientation dependence of the work function measured by FERP and TE methods. Although the displacement of the La atom 2.08 A normal to the (210) surface may seem rather large from the standpoint of a chemical bonding picture, such a large displacement is necessary to rationalize the fact that the work function of this plane is lower than the (100). A similar dipole contribution could be obtained with less displacement normal to the surface if both the first and second layer La atoms were displaced. Further experimental data is necessary to determine the rearrangement of the (210) surface. The dipole model overestimates the relative work function change between the various surfaces as shown in table 1. This effect has also been considered by Watson and Perlman [19], who note that surface ionic charge modification probably occurs in most polar crystals as this reduces the crystal capacitive potential energy. On this basis it is reasonable to consider a reduction in the assumed dipole charge of n = 1. For example, if n = 0.1 is used the predicted work function magnitude

differences

among

the various planes are of the same order of

as those measured.

In spite of the inability to properly specify the surface ionic charge, the dipole model does indicate that the orientation dependence of the surface dipole

moment,

as surmised

from geometrical

considerations,

plays an im-

portant role in determining the crystallographic variation of the work function. Most importantly, the dipole model does predict that the reconstructed (210) plane has a lower work function than the major low index planes. From the size of the net plane facets, the FIM pattern of the thermally annealed end form shown in fig. 3f indicates that the {llO}, {ill}, (112) and

59x

Fig.

6. Crystallographic

results

of this

map

of

the

i%$@$

%$2.40<+<2

;\\\\A

2.55

<+

2.65

<+<2.70eV

work

function

55eV < 2 65

distribution

for

eV

300 -c T < 1600

K based

on

the

study.

to some extent the { lOO} planes are the most thermodynamically stable planes. The absence of the (310) plane facets in fig. 3f may be interpreted that they are thermally unstable with respect to macroscopic, surface restructuring to lower index planes. At this juncture the main evidence that this is not so is the fact that no change FERP

work functions

in the high temperature were observed

thermionic

or room temperature

upon heating a macroscopic

(310) crystal

at 1800 K for 100 h.

5. Conclusions The crystallographic dependence of the work function of La$ is determined by the relative concentration of lanthanum in the outermost surface layer and the dipole moment per surface La atom. Surface reconstruction due to thermal equilibration and field evaporation can greatly alter lanthanum concentration and thus change the work function of specific planes. For example, the difference in I$ for the boron-rich field evaporated surface and lanthanum-rich thermally annealed surface is - 2.3 eV. Retarding potential and thermionic work function measurements on macroscopic single crystal show the (310) plane to have the lowest work function which is almost 0.2 eV lower than the

M. Gesley,L. W. Swanson / Low work function

planesofLaB,

599

(100) plane value. Based on this work, a crystallographic map of the work function distribution can be derived as shown in fig. 6. It is found that the relative order of the work function for various faces is +(310) < $(210) < +( 100) < +( 110) < +( 111) < $(211). This order can also be derived on the basis of the relative dipole layer contributions using the respective reconstructed surfaces. It has also been found that the field emission distribution is extremely sensitive to contaminants and requires careful experimental procedures to properly interpret the information obtained.

Acknowledgements

This work was supported in part by contract F19628-80-C-0117 from the Electronics System Division of Hanscom AFB and by grant NAG 3-434 from the NASA-Lewis Research Center. The authors are grateful to Drs. Paul Davis and Joe Hutta for helpful discussions during the course of this work and to Mr. Noel Martin who fabricated the LaB, single crystals.

References [I] L.W. Swanson, M.A. Gesley and P.R. Davis, Surface Sci. 107 (1981) 263. [2] C. Oshima, M. Aono, T. Tanaka, S. Kawai, R. Shimizu and H. Hagiwara, J. Appl. Phys. 51 (1980) 1201. [3] R. Nishitani. M. Aono, T. Tanaka, C. Oshima, S. Kawai, H. Iwasaki and S. Nakamura, Surface Sci. 93 (1980) 535. [4] M. Futamoto, S. Hosoki, H. Okano and U. Kawabe, J. Appl. Phys. 48 (1977) 3541. [5] C. Oshima, M. Aono, T. Tanaka, R. Nishitani and S. Kawai, J. Appl. Phys. 51 (1980) 997. [6] E. Gibson and J. Verhoeven, J. Phys. E8 (1975) 1003. [7] R. Strayer, W. Mackie and L. Swanson, Surface Sci. 34 (1973) 225. [S] M. Futamoto, T. Aita and U. Kawabe, Japan. J. Appl. Phys. 14 (1975) 1263. [9] E.W. Miiller and T.T. Tsong, Field Ion Microscopy (American Elsevier, New York, 1969). [lo] R. Shimizu, H. Onoda, H. Hagiwara and S. Ishii, J. Appl. Phys. 52 (1981) 6316. [ll] R. Shimizu. Y. Kataoka, T. Tanaka and S. Kawai, Japan. J. Appl. Phys. 14 (1975) 1089. [12] M. Futamoto and U. Kawabe, Surface Sci. 93 (1980) 1117. [13] K. Murakami, T. Adachi, T. Kuroda. S. Nakamura and 0. Komoda, Surface Sci. 124 (1983) 125. [14] E. Storms and B. Mueller, J. Phys. Chem. 82 (1978) 51. ]15] S. Nakamura. Y.S. Ng, T.T. Tsong and S.B. McLane, Jr., Surface Sci. 87 (1979) 656. [16] T. Sakurai, G. Robertson and Y. Kuk, Proc. Electrochem. Sot. 80-l (1980) 608. [17] L. Apker, E. Taft and J. Dickey, Phys. Rev. 74 (1948) 1462. [18] S.A. Chambers and L.W. Swanson, Surface Sci. 131 (1983) 385. [19] R.E. Watson and M.L. Perlman, Surface Sci. 122 (1982) 371.